MSDN Documentation

Microsoft Developer Network

Monitoring and Scaling Azure App Services

Effectively monitoring the health and performance of your Azure App Services, and configuring appropriate scaling strategies, are crucial for ensuring application availability, responsiveness, and cost-efficiency.

1. Monitoring Your App Service

Azure provides a comprehensive suite of tools within the Azure portal to monitor your App Services. Key metrics and features include:

Metrics

  • CPU Time: Percentage of time the CPU is busy.
  • Memory Working Set: Amount of physical memory used by the process.
  • HTTP Server Errors: Count of 5xx errors.
  • HTTP Average Response Time: Average time taken to respond to requests.
  • Data In/Out: Network traffic consumed.
  • Requests: Number of incoming requests.

Diagnostic Logs

Enable detailed logging to capture application-specific errors, request tracing, and other valuable information:

  • Application Logs: Log messages generated by your application code (e.g., exceptions, debug information).
  • Web Server Logs: Standard web server logs (e.g., IIS logs) providing details about each request.
  • Detailed Error Messages: Captures detailed error pages for HTTP errors.
  • Failed Request Tracing: Provides trace information for failed requests, helping to pinpoint the root cause.

Application Insights Integration

For deeper insights into application performance, integrate Azure Application Insights. It provides:

  • End-to-end request tracing.
  • Performance monitoring and dependency mapping.
  • Exception tracking and analysis.
  • Usage analytics.

To enable Application Insights: Navigate to your App Service in the Azure portal, go to "Application Insights" under "Settings", and click "Turn on Application Insights".

2. Scaling Your App Service

Scaling allows you to adjust the resources allocated to your App Service to meet demand. Azure App Services support two primary scaling methods:

Manual Scaling

You can manually set the number of instances or the size of the instance. This is useful for predictable workloads or for testing specific scaling configurations.

To configure manual scaling: Navigate to your App Service, go to "Scale out (App Service plan)" under "Settings", and select "Manual scale". Choose the desired number of instances.

Automatic Scaling (Autoscaling)

Autoscaling adjusts the number of instances automatically based on predefined rules and metrics. This is ideal for variable or unpredictable workloads.

To configure autoscaling: Navigate to your App Service, go to "Scale out (App Service plan)" under "Settings", and select "Scale out automatically".

Configure the following:

  • Default Count: The number of instances to scale to if no rules are met.
  • Minimum/Maximum Instances: The bounds for autoscaling.
  • Scale Rules: Define conditions based on metrics (e.g., CPU percentage, queue length) that trigger an increase or decrease in instances.

Autoscaling Rules Examples

Here are some common autoscaling rules:

// Example: Scale out when CPU percentage is above 70% for 10 minutes
{
  "name": "ScaleOutCpuHigh",
  "description": "Scale out when CPU percentage is above 70%",
  "change_count": 1,
  "direction": "Increase",
  "type": "Metric",
  "metric_name": "CpuPercentage",
  "metric_namespace": "Microsoft.Web/serverFarms",
  "statistic": "Average",
  "time_grain": "PT10M", // 10 minutes
  "time_window": "PT10M", // Evaluate over 10 minutes
  "threshold": 70
}

// Example: Scale in when CPU percentage is below 30% for 15 minutes
{
  "name": "ScaleInCpuLow",
  "description": "Scale in when CPU percentage is below 30%",
  "change_count": -1,
  "direction": "Decrease",
  "type": "Metric",
  "metric_name": "CpuPercentage",
  "metric_namespace": "Microsoft.Web/serverFarms",
  "statistic": "Average",
  "time_grain": "PT15M", // 15 minutes
  "time_window": "PT15M", // Evaluate over 15 minutes
  "threshold": 30
}
Best Practice: Start with a conservative autoscaling configuration and gradually adjust rules based on observed performance and cost. Monitor your application closely after implementing autoscaling.

3. Setting Up Alerts

Beyond basic monitoring, set up alerts to proactively notify you of potential issues.

  • Configure alerts based on critical metrics (e.g., high CPU, increased error rates, low disk space).
  • Set up alert actions to send emails, trigger webhooks, or run automation runbooks.

To configure alerts: In the Azure portal, navigate to your App Service, go to "Alerts" under "Monitoring", and click "+ Create alert rule".

Tip: Consider using Azure Advisor for recommendations on optimizing performance, security, and cost for your App Services.